Local Conformational Changes in the Catalytic Core of the Trans

Sep 14, 2002 - The hepatitis delta virus ribozyme is among a class of small endonucleolytic RNAs that catalyze a reversible self- cleavage reaction ne...
0 downloads 13 Views 247KB Size
Download PDF
Biochemistry 2002, 41, 12051-12061

12051

Local Conformational Changes in the Catalytic Core of the Trans-Acting Hepatitis Delta Virus Ribozyme Accompany Catalysis† Dinari A. Harris, David Rueda, and Nils G. Walter* Department of Chemistry, The UniVersity of Michigan, 930 North UniVersity, Ann Arbor, Michigan 48109-1055 ReceiVed May 8, 2002; ReVised Manuscript ReceiVed August 15, 2002

ABSTRACT: The hepatitis delta virus (HDV) is a human pathogen and satellite RNA of the hepatitis B virus. It utilizes a self-cleaving catalytic RNA motif to process multimeric intermediates in the doublerolling circle replication of its genome. Previous kinetic analyses have suggested that a particular cytosine residue (C75) with a pKa close to neutrality acts as a general acid or base in cleavage chemistry. The crystal structure of the product form of a cis-acting HDV ribozyme shows this residue positioned close to the 5′-OH leaving group of the reaction by a trefoil turn in the RNA backbone. By modifying G76 of the trefoil turn of a synthetic trans-cleaving HDV ribozyme to the fluorescent 2-aminopurine (AP), we can directly monitor local conformational changes in the catalytic core. In the ribozyme-substrate complex (precursor), AP fluorescence is strongly quenched, suggesting that AP76 is stacked with other bases and that the trefoil turn is not formed. In contrast, formation of the product complex upon substrate cleavage or direct product binding results in a significant increase in fluorescence, consistent with AP76 becoming unstacked and solvent-exposed as evidenced in the trefoil turn. Using AP fluorescence and fluorescence resonance energy transfer (FRET) in concert, we demonstrate that this local conformational change in the trefoil turn is kinetically coincidental with a previously observed global structural change of the ribozyme. Our data show that, at least in the trans-acting HDV ribozyme, C75 becomes positioned for reaction chemistry only along the trajectory from precursor to product.

The hepatitis delta virus ribozyme is among a class of small endonucleolytic RNAs that catalyze a reversible selfcleavage reaction necessary for the replication and propagation of their satellite RNA genomes. Specifically, the hepatitis delta virus ribozyme is a unique RNA motif found in the human hepatitis delta virus (HDV)1 (1). HDV is a satellite of the hepatitis B virus (HBV); coinfection of HDV and HBV results in intensification of the disease symptoms associated with the hepatitis B virus (2). The small RNA genome of HDV replicates through a double-rolling circle mechanism, whereby multimeric units of genomic and antigenomic RNA strands are produced, followed by self-cleavage and ligation into circular monomers (1, 3). Self-cleavage activity in the genomic and antigenomic RNAs resides within continuous 85-nucleotide sequences that both form a nearly identical secondary structure consisting of a nested double pseudoknot (4, 5). The genomic and antigenomic forms of the HDV ribozyme catalyze self-cleavage by a transesterification reaction, which requires deprotonation of the adjacent 2′-OH group and its nucleophilic attack on the scissile phosphate, resulting in formation of 2′,3′-cyclic phosphate and 5′-OH termini (5). † This work was supported by NIH Grant GM62357 to N.G.W., a Rackham Merit predoctoral fellowship and NIH Molecular Biophysics Training Grant to D.A.H., and a postdoctoral fellowship from the Swiss National Fonds to D.R. * To whom correspondence should be addressed. Phone: (734) 6152060. Fax: (734) 647-4865. E-mail: [email protected]. 1 Abbreviations: AP, 2-aminopurine; FRET, fluorescence resonance energy transfer; fwhm, full width at half-maximum; HDV, hepatitis delta virus.

The reaction mechanism of the HDV ribozyme has been extensively studied. The crystal structure of the self-cleaved genomic ribozyme reveals that the base cytosine 75 (C75) is situated in the active site cleft and, thus, in the proximity of the 5′-OH leaving group (Figure 1a,b). Therefore, C75 in the genomic ribozyme has been proposed to participate directly in reaction chemistry as either a general acid or general base catalyst (6). Several biochemical and mutagenesis studies support the idea that C75 in the genomic ribozyme and the corresponding RC76 (R used to distinguish antigenomic numbering) in the antigenomic ribozyme are involved in catalysis (7-10). The pH dependence of self-cleavage (or cis cleavage) by the HDV ribozyme reveals a macroscopic apparent pKa that approaches neutrality. In a widely accepted model, this pKa reflects the ionization equilibrium of N3 in C75 which therefore is strongly shifted in the folded ribozyme compared to that in the free base (pKa ≈ 4.2). A decrease in this pKa for selfcleavage of an antigenomic ribozyme with an RC76A mutation was observed, consistent with A substituting for C in this position to act as a general base catalyst (8). However, the pH profile of the genomic ribozyme in the presence of 1 M NaCl and 1-100 mM EDTA favors a model where C75 acts as a general acid during catalysis (9, 11). This latter mechanism is in agreement with the crystal structure of the self-cleaved genomic ribozyme, which shows N3 of C75 within hydrogen bonding distance of the 5′-OH leaving group. In addition, C75 is hydrogen bonded to the phosphate group of C22, increasing the local electron density and providing a mechanism for a shift in pKa, such that it is

10.1021/bi026101m CCC: $22.00 © 2002 American Chemical Society Published on Web 09/14/2002

12052 Biochemistry, Vol. 41, No. 40, 2002

FIGURE 1: Sequence and structure of the HDV ribozyme. (a) Twodimensional representation of our synthetic trans-acting HDV ribozyme construct D1. The ribozyme-3′-product complex is shown annotated according to ref 66 with the tertiary structure of the postcleavage crystal structure which it models (6). The colored nucleotides are the ones shown in detail in panel b. The ribozyme portion consists of two separate RNA strands (A and B). Substrate strand S3 (outlined) is cleaved at the arrow into the 5′- and 3′products (5′P and 3′P, respectively), and 5′P dissociates from the ribozyme-3′-product complex. (b) Trefoil turn in J4/2 as observed in the postcleavage crystal structure (red). C75 is positioned deep in the catalytic core making a single hydrogen bond (orange) with the 5′-hydroxyl leaving group of G1 (yellow), whereas G76 loops out exposing it to solvent. Also shown as an overlay (gray) is the more regular backbone trajectory between G76 and U79. This figure was generated using VMD 1.7.2 and Pov-Ray 3.1g for Windows.

protonated around neutral pH (6). Thus, C75 is in a position in which it may donate a proton to relieve the developing charge on the leaving 5′-oxyanion group in the reaction transition state (6). Recently, nucleotide analogue interference mapping (NAIM) of the genomic ribozyme has provided further evidence for the importance of the ionization of C75 in the cis cleavage reaction, as it is the only cytidine residue that must be protonated for ribozyme activity (10). The use of an RNA base to facilitate transition-state chemistry appears to be surprisingly common among RNA catalysts. For example, biochemical and structural studies suggest that the hairpin ribozyme may use two catalytically essential nucleotide residues, G8 and A38, to function as the proton acceptor and donor, respectively (12, 13). An

Harris et al. analogous proposal has been made for the large subunit of the ribosome, where no amino acid side chains or metal ions but rather a conserved adenosine (A2451 in Escherichia coli numbering) is found in sufficient proximity to potentially act as a catalyst in peptidyl transfer (14), although this is still controversial (15, 16). These discoveries reveal that RNA bases in ribozymes can participate in catalysis in a manner similar to that of amino acid side chains in protein enzymes. Hence, the finding that the HDV ribozyme utilizes one of its RNA side chains in acid-base catalysis has led the way to fundamental insights into RNA catalysis. Studies of the folding pathway of the HDV ribozyme are key to our understanding of how the RNA positions C75 in the active site cleft such that it becomes both chemically activated and poised for catalysis. The crystal structure of the self-cleaved genomic ribozyme gives hints about how this positioning occurs. C75 and its two adjacent nucleotides, G76 and G74, form a “trefoil” turn that pushes C75 deeply into the catalytic core (Figure 1b) (6). In particular, extrusion of G76 allows for a backbone geometry that projects C75 toward the 5′-OH leaving group. With this in mind, we have devised a fluorescence-based technique for monitoring the unstacking of the nucleotide in position 76 as it extrudes from neighboring bases into solvent, by substituting 2-aminopurine (AP) for G76. AP is a fluorescent guanine analogue that is highly sensitive to local stacking interactions with other bases and can be selectively excited in the presence of the natural bases (17). Mutagenesis studies suggest that the chosen substitution site can be mutated with only a modest loss of activity (7). When monitoring conformational changes at position 76 through AP fluorescence in the trans-acting HDV ribozyme, we observe a substantial local conformational change that accompanies ribozyme catalysis. Our results complement a previous study in which we found, using fluorescence resonance energy transfer (FRET), that a significant global conformational change concurs with catalysis (18). Using AP fluorescence and FRET in concert, we demonstrate here that the global structural change upon cleavage is coupled to, and probably the consequence of, local conformational rearrangements in and around the catalytic core that lead to proper positioning of C75. MATERIALS AND METHODS Preparation of RNA Oligonucleotides. All RNA oligonucleotides were obtained commercially from Dharmacon Research, Inc., with 2′-protection groups (19) and were deprotected following the manufacturer’s recommendations. Deprotected RNA was purified by denaturing 20% polyacrylamide, 8 M urea, gel electrophoresis, diffusion elution into 0.5 M NH4OAc, 0.1% SDS, and 0.1 mM EDTA overnight at 4 °C, chloroform extraction, and ethanol precipitation, followed by C8 reverse-phase HPLC with a linear acetonitrile elution gradient in triethylammonium acetate, as described previously (20). The noncleavable substrate analogue was modified with a 2′-methoxy group at the cleavage site (Figure 1a). The 3′-product had the sequence 5′-GGGUCGG-3′. RNA concentrations were calculated from the absorbance at 260 nm after background correction. 5′-Fluorescein was incorporated into the RNA during synthesis by Dharmacon Research, Inc., while tetramethylrhodamine was postsynthetically attached to a 3′amino linker as previously described (18). The concentrations

Local Conformational Changes of the HDV Ribozyme of fluorescein- and tetramethylrhodamine-labeled RNAs were corrected for the additional absorbance of the fluorophores by using the relations A260/A492 ) 0.3 and A260/A554 ) 0.49, respectively. CleaVage Reactions. Activities of ribozymes with G76 modifications were determined using the three-strand HDV ribozyme construct depicted in Figure 1a. All cleavage reactions were conducted under single-turnover (pre-steadystate) conditions and using standard conditions of 40 mM Tris-HCl (pH 7.5) and 11 mM MgCl2, at 37 °C, unless otherwise stated. The ribozyme was prepared by preannealing strand A and twice the concentration of strand B in standard buffer, heating to 70 °C for 2 min, and cooling to room temperature. After preincubation at 37 °C for 15 min, a trace amount (40000 peak counts, and under magic angle polarizer conditions. To measure donor-acceptor distances, two timeresolved fluorescence decays were collected, with and without the acceptor in place. The effect of the acceptor on the decay of fluorescein emission in the doubly labeled complex was then used to extract a three-dimensional Gaussian distance distribution between the two fluorophores as previously described in detail (18, 25, 26). To calculate a mean distance, a value of 55 Å for the Fo¨rster distance R0 of fluorescein and tetramethylrhodamine was used (25), assuming a value of 2/3 for the orientation factor as experimentally supported by low fluorophore anisotropies (18). FRET Gel Shift Assays. FRET gel shift assays were conducted to test the homogeneity of the various ribozymesubstrate complexes, as previously described (18). Nondenaturing 10% polyacrylamide (19:1 acrylamide:bisacrylamide ratio) gels containing 40 mM Tris-HCl (pH 7.5) and 11 mM Mg(OAc)2 were assembled with the electrophoresis unit and equilibrated to 4 °C for at least 2 h. Ten picomoles of doubly fluorophore-labeled ribozyme strand B was annealed to 20 pmol of strand A by heating for 2 min to 70 °C and cooling to room temperature in 40 mM Tris-HCl (pH 7.5), 11 mM MgCl2, and 10% glycerol. The ribozyme was equilibrated at 37 °C for at least 15 min prior to addition of 50 pmol of substrate, noncleavable substrate analogue, or 3′-product (total volume of 20 µL). These samples were loaded on the gel, and an electric field of 5 V/cm was immediately applied. After electrophoresis for 24 h, the gel was scanned between its low-fluorescence glass plates in a FluorImager SI fluorescence scanner with ImageQuant software (Molecular Dynamics) as described previously (18, 27). A laser excites fluorescein at 488 nm, and the gel is scanned for fluorescence emission using a photomultiplier tube with either a 530 nm band-pass (for the donor fluorescein) or a 610 nm long-pass filter (for the acceptor tetramethylrhodamine). RNAs labeled with only fluorescein and only tetramethylrhodamine were included as color calibration standards. From the volume reports, a measure of FRET efficiency of selected bands was calculated as Facceptor/Fdonor. With the readout of Fdonor defined as being green and that of Facceptor as red, the corresponding color images were superimposed using Photoshop 5.5 (Adobe) to generate Figure 6. RESULTS A Base Substitution of G76 with AP Is Tolerated in a TransActing HDV Ribozyme with an Only Modest Decrease in Catalytic ActiVity. The trans-cleaving HDV ribozyme used in this study, D1, is a synthetic three-strand construct, which consists of ribozyme strands A and B and substrate strand S3 (Figure 1a). Previously, we characterized in detail the catalytic activity of this trans-cleaving construct under a variety of conditions (18). Our data showed that this construct has cleavage rate constants as well as metal ion, temperature, and pH dependencies similar to those of other trans-acting HDV ribozymes. To evaluate the effect that a G76AP

Harris et al.

FIGURE 2: Cleavage of substrate strand S3 by HDV ribozyme construct D1-AP76 under single-turnover conditions. (a) Cleavage time courses of D1-AP76 under standard conditions of 40 mM TrisHCl (pH 7.5) and 11 mM MgCl2 at 37 °C, with varying ribozyme concentrations [(O) 50, (0) 100, (]) 200, (4) 400, (crossed box) 800, (K) 1200, and (tilted triangle) 1600 nM]. Data were fit to a single-exponential increase function (s) to yield the rate constants kobs reported in the inset (total fraction cleaved, ∼70%). In the inset, the dependence of kobs on ribozyme concentration was fit to a binding equation (s) (see Materials and Methods), yielding a kcleav of 0.20 min-1 (KM ) 114 nM). (b) pH dependence of the observed cleavage rate constants. The data were fit [see Materials and Methods; D1 (s9s) and D1-AP76 (- -O- -)] to yield the two pKa values for each construct reported in Table 1.

mutation has on this construct, we measured its cleavage rate constants, as well as its metal ion and pH dependencies, and compared them to those of the D1 wild type with G76. First, we measured the cleavage rate constant under standard single-turnover (pre-steady-state) reaction conditions, which consisted of trace amounts of radiolabeled S3 substrate with a saturating excess of 400 nM ribozyme in 40 mM Tris-HCl (pH 7.5) and 11 mM MgCl2 at 37 °C (see Materials and Methods). In addition, reactions were performed at ribozyme concentrations varying between 25 and 1800 nM to ensure reaching saturation. Figure 2a shows the resulting reaction time courses for the G76AP mutant. The observed pseudo-first-order rate constants (kobs) were plotted as a function of ribozyme concentration and fit to a simple binding equation (inset of Figure 2a; see Materials and Methods), yielding rate constants for the rate-limiting step of cleavage (kcleav) at 37 °C of 1.34 and 0.20 min-1 for the D1 wild type and the G76AP mutant, respectively (Table 1). The ∼7-fold decrease in catalytic activity with the G76AP mutation is on the same order of magnitude as that for other mutations, including G76 to adenine (A), inosine (I), uracil (U), or 4-thiouracil (SU) (Table 2). Cleavage rate constants followed this trend: wild type > G76SU ≈ G76I > G76U > G76AP ≈ G76A. A moderate decrease in cleavage activity upon substitution of G76 is also consistent with previous mutagenesis studies, particularly in the cis-cleaving genomic HDV ribozyme where substituting G76 for U results in an ∼15-fold decrease in catalytic activity (7).

Local Conformational Changes of the HDV Ribozyme Table 1: Cleavage Rate Constants, Metal Ion Dissociation Constants, and pKas for HDV Ribozyme Construct D1 and Its G76AP Mutanta parameter -1

kcleav (min ) (pH 7.5, 11 mM Mg2+, 37 °C) Mg1/2 (mM) pKa1, pKa2 kon (M-1 min-1) koff (min-1)

D1-S3

D1-AP76-S3

1.34 ( 0.07

0.20 ( 0.01

9(1 5.7, 8.6 ND ND

17 ( 3 5.1, 8.7 (2.4 ( 0.2) × 106 0.20 ( 0.04

a The single-turnover cleavage rate constant k cleav was determined with substrate strand S3 and saturating ribozyme as described in the text (standard conditions, 40 mM Tris-HCl, pH 7.5, 11 mM MgCl2, and 37 °C). Mg1/2s were derived as described in Materials and Methods, pKas from the data in Figure 2b, and kon and koff from the data in panels b and c of Figure 3, respectively. ND means not determined.

Table 2: Cleavage Activity of G76 Modifications on the HDV Ribozymea kobs (min-1) wild type SU I

0.84 ( 0.02 0.54 ( 0.02 0.53 ( 0.01

kobs (min-1) U AP A

0.28 ( 0.04 0.17 ( 0.01 0.17 ( 0.01

a Observed rate constants (k ) were measured under standard obs conditions (single-turnover, 40 mM Tris-HCl, pH 7.5, 11 mM MgCl2, 400 nM ribozyme, and 37 °C) for trans-cleaving HDV ribozyme construct D1 (Figure 1a); errors were estimated from at least three independent measurements.

Next, we compared Mg2+ binding affinities of the wildtype and G76AP mutant D1 constructs. Cleavage activity was measured under standard single-turnover conditions (at 400 nM ribozyme excess) but at varying magnesium concentrations (2-200 mM Mg2+). The observed rate constants for the G76 wild-type ribozyme increase substantially from 0.075 to 1.47 min-1 in this range, while the G76AP mutant shows a more modest increase from 0.26 and 0.393 min-1. The resultant magnesium titration midpoints (Mg1/2) are 9 and 17 mM, respectively (Table 1). These Mg2+ affinities are somewhat lower than that reported for a cis-cleaving HDV ribozyme (1 mM) (28), but higher than that of another threestrand ribozyme construct (40 mM) (29). Finally, we compared the pH dependence of cleavage of the wild type and the G76AP mutant to verify that the latter utilizes a mechanism similar to that of the unmodified D1 construct. Such pH profiles have been critical in establishing the involvement of C75 in the catalytic mechanism of the HDV ribozyme. Figure 2b shows the resulting curves. Both constructs show bell-shaped pH profiles similar to those of previously characterized trans-acting (22) and cis-acting HDV ribozymes (30). From the pH dependence of the wildtype G76 D1 construct, two pKas of 5.7 and 8.6 are derived, while the pKas of the G76AP mutant are 5.1 and 8.7 (Figure 2b and Table 1). It is noteworthy that both pH profiles suggest that a pH-dependent step is rate-limiting at low pH (5, 31). Steady-State 2-Aminopurine Fluorescence Assays ReVeal Differences between Pre- and PostcleaVage Structures and Report on the Folding Kinetics of the HDV Ribozyme. AP is a strongly fluorescent base (excitation maximum at 320 nm, emission maximum at 360 nm, quantum yield of 68%) and is highly sensitive to local stacking interactions with other bases. The fluorescence of AP is quenched when it stacks with other bases, but increases substantially when it

Biochemistry, Vol. 41, No. 40, 2002 12055 becomes fully exposed to solvent (32-34). AP has recently been utilized as a fluorescent probe for local dynamics in other catalytic RNAs, such as the hammerhead and hairpin ribozymes (12, 23, 24). To observe localized structural changes in the catalytic core of the HDV ribozyme, we have monitored changes in AP steady-state fluorescence of our G76AP modified D1 construct (D1-AP76) upon addition of either substrate or product. In all fluorescence assays, saturating 2- and 5-fold excesses of the unmodified strand A and substrate (or product), respectively, were added under standard conditions (see Materials and Methods). The fluorescence changes of AP76 indicate that a substantial conformational rearrangement around this position accompanies catalysis (Figure 3a). In particular, addition of saturating concentrations of a chemically blocked, noncleavable substrate analogue of S3 (ncS3) to the assembled ribozyme results in a slight (∼12%) decrease in relative AP fluorescence (inset of Figure 3a). This fluorescence quenching suggests that AP becomes more stacked as a result of formation of the ribozyme-substrate (precursor) complex. The pseudo-first-order rate constant for this decrease is linearly dependent on the excess concentration of ncS3, indicating that this fluorescence decrease is a direct result of substrate binding. On the basis of this linear dependence, we deduce a second-order substrate binding rate constant kon of 2.4 × 106 M-1 min-1 (Figure 3b and Table 1), similar to a previously determined value for this ribozyme-substrate complex at 25 °C instead of 37 °C (18). In contrast, when a saturating concentration of the cleavable substrate S3 is added, we observe an ∼7-fold increase in AP fluorescence (Figure 3a). The rate constant of 0.27 min-1 for this single-exponential increase is similar to the cleavage rate constant obtained in the standard autoradiographic assay (0.17 min-1; see above). This suggests that the observed fluorescence increase is the result of substrate cleavage, which leads to rapid dissociation of its 5′-portion (5, 18) and formation of the ribozyme-3′-product complex. Notably, initial binding of substrate, which is associated with a fluorescence decrease, results in a slight lag phase in the fluorescence increase (inset of Figure 3a). Direct formation of the ribozyme-3′-product complex was initiated by adding a saturating excess of the 3′-product of the substrate (termed 3′P, sequence of 5′-GGGUCGG-3′) to the assembled ribozyme. Previously, it has been shown that the 3′-product remains bound to the ribozyme after cleavage, while the 5′-product rapidly dissociates (4, 18). We find that addition of 3′P and formation of the ribozyme-3′P complex result in a fast (slightly double-exponential) ∼9.5-fold increase in the relative AP76 fluorescence. We estimate the rate constant for this fluorescence increase to be ∼7.1 min-1 at 1 µM 3′P. This substantial dequenching of its fluorescence implies that AP76 becomes significantly unstacked and exposed to solvent in the ribozyme-3′-product complex. This observation is consistent with the postcleavage crystal structure of the genomic form of the cis-cleaving HDV ribozyme, which shows G76 extruding into solvent (Figure 1b). Furthermore, the fluorescence increase following substrate cleavage is approximately 75% of that which occurs upon direct formation of the ribozyme-3′P complex, consistent with the 75% conversion of S3 into 3′P by the G76AP mutant under standard cleavage conditions (Figure 2a).

12056 Biochemistry, Vol. 41, No. 40, 2002

Harris et al.

FIGURE 4: Fluorescence of 2-aminopurine 2′-deoxyribose 5′triphosphate (dAPTP) and 2-aminopurine in position 76 of the D1 ribozyme construct and its complexes as a function of pH (see Materials and Methods). The wavelength of excitation was 290 nm, and the emission spectra were monitored at the AP peak at 360 nm. pKa values of 3.9 and 4.8 for the free nucleoside triphosphate and the ribozyme-3′P complex, respectively, were derived by fitting the data to the equation Fobs ) Fmax/(1 + 10pKa-pH).

FIGURE 3: Steady-state 2-aminopurine fluorescence measurements using 200 nM ribozyme D1-AP76 in 40 mM Tris-HCl (pH 7.5) and 11 mM MgCl2 at 37 °C. (a) Change over time in the relative AP fluorescence of D1-AP76 ribozyme upon addition of (saturating) 1000 nM noncleavable substrate analogue (ncS3), cleavable substrate (S3), or 3′-product (3′P), as indicated. The data set for ncS3 was fit with a single-exponential decrease function to yield a rate constant of 1.0 min-1 that is reported in panel b. The data set for S3 gave a single-exponential increase that pertains to substrate cleavage with a rate constant of 0.27 min-1. The 3′P data set increases double-exponentially with a fast-phase rate constant of 7.1 min-1 (slow phase, 0.1 min-1 with 12% of the fast-phase amplitude). In the inset, rescaling of the early data points highlights the slight decrease in AP fluorescence upon formation of the ribozyme-ncS3 complex (O) and the slight lag phase, due to substrate binding, in the fluorescence increase upon cleavage of the S3 substrate (9) (gray line, exponential fit to the increase alone). (b) Concentration dependence of the observed pseudo-first-order rate constant upon addition of excess ncS3 to the ribozyme. The slope of the linear regression line yields the bimolecular binding rate constant kon. (c) Dissociation of the noncleavable substrate analogue from its ribozyme complex. The ribozyme-ncS3 complex (200 nM, with a 1000 nM excess of ncS3) was chased with 4 µM 3′-product, and the resulting increase in 2-aminopurine fluorescence was fit to a single-exponential increase function to yield a dissociation rate constant koff at 0.20 min-1 (s). There is no significant dependence on the chase concentration (inset).

Next, we exploited the fact that 3′P has a higher affinity for binding to the HDV ribozyme than the substrate (4, 18), providing an avenue for assessing substrate dissociation. To this end, the ribozyme-ncS3 precursor complex was formed and subsequently chased with a 4 µM excess of 3′P. This resulted in a single-exponential increase in AP76 fluorescence due to dissociation of the substrate and replacement with 3′P chase (Figure 3c). The observed rate constant koff of 0.20

min-1 is independent of the concentration of chase (inset of Figure 3c). The substrate dissociation and binding rate constants, koff and kon, respectively, measured at 37 °C, define the equilibrium dissociation constant KD ()koff/kon) which equals 110 nM at 37 °C, 3-fold higher than a previously determined value at 25 °C (18), as expected for a shift to a higher temperature. The 2-Aminopurine Fluorescence as a Function of pH ReVeals Significant Differences in the Chemical EnVironment of the Catalytic Core before and after CleaVage. Our steadystate AP76 fluorescence assays indicate that there are structural differences between the precursor and postcleavage conformations near the catalytic core of the HDV ribozyme. The AP fluorescence in the ribozyme and ribozyme-ncS3 complex is significantly quenched, presumably as a result of local stacking interactions. In contrast, significant unquenching (i.e., unstacking) of AP76 is observed upon formation of the 3′-product complex (Figure 3a). On the basis of these observations, we decided to monitor the protonation state of the N1 position of AP76 in the ribozyme and its substrate and 3′-product complexes by measuring the fluorescence intensity as a function of pH. At low pH, the relative fluorescence intensity of 2-aminopurine is reduced as a result of N1 protonation (17, 35). First, we measured the pH titration curve of 2-aminopurine 2′-deoxyribose 5′-triphosphate in solution as a control, which resulted in a pKa of 3.9 (Figure 4), consistent with a previously reported value of 3.8 (17). Next, we measured the pH titration curve of the AP76 in the ribozyme, the ribozyme-ncS3 complex, and the ribozyme-3′P complex. These curves indicate clear differences in titration of AP76 that relate to its local environment (Figure 4). From the pH titration curve of the ribozyme-3′ product complex an AP76 pKa of 4.8 is derived. Although this experimentally determined pKa is nearly 1 pH unit higher than that of free AP in solution, the AP76 fluorescence intensity in this complex essentially titrates as the free base. This result further supports the idea that position 76 in the product complex is looped out and exposed to solvent. Conversely, the fluorescence in the ribozyme and the ribozyme-ncS3 complex is strongly quenched throughout the whole pH range. Although the low intensity of AP76 fluorescence in these constructs impedes extraction of reliable pKas from the titration curves, these curves show that the local environments in the free ribozyme and the ribozyme-

Local Conformational Changes of the HDV Ribozyme ncS3 complex are very different from that in the ribozyme3′P complex. These observations are consistent with the notion that significant local structural changes around the catalytic core occur upon cleavage in the HDV ribozyme. Synchronous 2-Aminopurine Fluorescence and SteadyState FRET Kinetics Suggest that Local and Global Conformational Changes Occur Simultaneously. Previously, we have shown by fluorescence resonance energy transfer (FRET) that this trans-acting HDV ribozyme undergoes a global conformational change upon substrate cleavage (18). However, these experiments could not reveal whether this global conformational change is accompanied by local structural rearrangements. In light of our success with AP76 as a local conformational probe near the catalytic core of the HDV ribozyme, we sought to test whether the local and global conformational changes are coupled. To this end, we utilized AP76 and FRET in concert, by employing a modified strand B of our D1 construct with the G76AP substitution and terminal 5′-fluorescein and 3′-tetramethylrhodamine labels as donor-acceptor FRET pair (Figure 1a). To ensure that all fluorophore-labeled strands are converted into complexes, we added 2- and 5-fold excesses of the unmodified strand A and substrate (or product), respectively (see Materials and Methods). Similar results were obtained when AP76 and the FRET fluorophores were incorporated separately into strands B and A, respectively, and used at equimolar concentrations (data not shown). Figure 5a shows the changes in relative AP76 fluorescence and FRET efficiency upon addition of a 1000 nM excess of noncleavable substrate analogue, ncS3, to the assembled AP76 and FRET-labeled ribozyme. The synchronous decrease in both signals was fit to single exponentials that yield rate constants of 1.3 and 1.4 min-1, respectively. These values compare well with the rate constant for the AP76 fluorescence decrease under identical conditions, but in the absence of the FRET probes (1.0 min-1, Figure 3a). The AP fluorescence and FRET changes after addition of a saturating concentration of 1000 nM of either cleavable substrate S3 or 3′P to the ribozyme show characteristics expected for formation of the ribozyme-3′-product complex. Specifically, when adding a saturating concentration of S3, we observed an increase in the relative AP fluorescence and a decrease in the relative FRET efficiency (Figure 5b). The two fluorescence changes are again synchronous and, when fit to single-exponential increase and decrease functions, respectively, yield rate constants of 0.21 and 0.24 min-1, respectively. These fluorescence changes are indicative of substrate cleavage and their rate constants comparable to the ones obtained from the radioactive cleavage assay (0.17 min-1) and monitoring of AP76 fluorescence in the absence of FRET fluorophores (0.27 min-1). Finally, when we directly initiated formation of the ribozyme-3′-product complex by adding 3′P to the ribozyme, a substantial increase in the relative AP fluorescence and a synchronous decrease in relative FRET efficiency were observed (Figure 5c), with rate constants of 5.2 and 5.7 min-1, respectively. These values are in reasonable agreement with the value obtained from AP76 probing in the absence of FRET probes (7.1 min-1). These results demonstrate that local conformational changes of the trans-acting HDV ribozyme as monitored by AP76 fluorescence occur simul-

Biochemistry, Vol. 41, No. 40, 2002 12057

FIGURE 5: Time courses of the relative AP fluorescence (O) and FRET efficiency (9) of the doubly fluorescein- and tetramethylrhodamine-labeled D1-AP76 ribozyme upon addition of 1000 nM noncleavable substrate analogue (ncS3), cleavable substrate (S3), or 3′-product (3′P) in 40 mM Tris-HCl (pH 7.5), 11 mM MgCl2, and 25 mM DTT at 37 °C. (a) Formation of the ribozyme-ncS3 complex results in slight decreases in relative AP fluorescence and FRET efficiency with rate constants of 1.3 and 1.4 min-1, respectively. (b) The data sets for S3 pertain to substrate cleavage and give a single-exponential increase in relative AP fluorescence and a concurrent decrease in relative FRET efficiency with rate constants of 0.21 and 0.24 min-1, respectively. (c) Direct formation of the product complex by the addition of 3′P gives a fast doubleexponential increase in relative AP fluorescence and a concurrent decrease in FRET efficiency, with fast-phase rate constants of 5.2 and 5.7 min-1, respectively. The slow phases contribute 20 and 11% of their respective fast phases, with rate constants of 0.1 min-1 for both the AP fluorescence increase and FRET decrease. In the absence of further studies of their nature, we cannot distinguish whether these phases are due to a small fraction of a second RNA species with slower folding kinetics, or due to a slow second kinetic step in folding of the main RNA species. In either case, the minor slow components do not change our conclusions about the coincidence of local and global folding events as they are present in both the AP and FRET data sets.

taneously with the global conformational changes observed by FRET. A FRET Gel Shift Assay Demonstrates the Homogeneity of the Various, Structurally Distinct Ribozyme Complexes. We utilized a previously described FRET gel mobility assay (18) to examine the homogeneity of our various ribozyme complexes. To study the effect that the G76AP substitution has on the overall folding of the ribozyme and its complexes, the doubly labeled G76AP mutant was electrophoresed on a

12058 Biochemistry, Vol. 41, No. 40, 2002

Harris et al. Table 3: Time-Resolved FRET Analysis of the Doubly Fluoresceinand Tetramethylrhodamine-Labeled D1-AP76 Ribozyme and Its Complexesa parameter

D1-AP76

D1-AP76-ncS3

D1-AP76-3′P

mean donor-acceptor distance (Å) fwhm (Å) χ2

48 ( 1

52 ( 1

65 ( 1

25 ( 1 1.1

21 ( 1 1.2

33 ( 1 1.1

a Donor-acceptor distances and their full widths at half-maximum (fwhm) were measured as described in Materials and Methods in 40 mM Tris-HCl (pH 7.5), 11 mM MgCl2, and 25 mM DTT at 37 °C; errors were estimated from two independent measurements. χ2 is the reduced χ2 value.

FIGURE 6: Nondenaturing gel electrophoresis of doubly fluoresceinand tetramethylrhodamine-labeled D1-AP76 complexes (see Materials and Methods). The free AP76 mutant ribozyme (D1-AP76, lane 1) comigrates with its ribozyme-3′-product complex (D1AP76+3′P, lane 2, top panel), but can be distinguished by the ratio of acceptor to donor fluorescence (bottom panel). The ribozymesubstrate complex (D1-AP76+S3, lane 3) undergoes catalysis and is observed as the ribozyme-3′-product complex. The ribozymenoncleavable substrate analogue complex (D1-AP76+ncS3, lane 4) stays intact and migrates slower.

nondenaturing polyacrylamide gel alone and in complex with S3, ncS3, and 3′P. The relative FRET efficiencies for all bands in the gel were derived from a quantitative Fluorimager analysis (Figure 6) (see Materials and Methods). We found that the ribozyme alone as well as its complexes migrate as homogeneous populations, with less than 15% of the overall fluorescence intensity resulting from alternate bands (data not shown). The FRET efficiencies of the gel shift bands of the ribozyme and its complexes containing the G76AP substitution are consistent with our steady-state fluorescence assays of this mutant. Lane 1, loaded with ribozyme alone, shows a band with a high FRET ratio (“or red-shifted”, i.e., acceptor fluorescence-dominated band). Lane 2, loaded with ribozyme-3′P complex, shows a band that comigrates with the ribozyme alone and has a significantly decreased FRET ratio (“green-shifted”). Lane 3, loaded with the ribozymecleavable substrate complex, shows a band that comigrates with the ribozyme-3′P complex, which is the result of substrate cleavage and formation of the ribozyme-3′P complex. Lane 4, loaded with the ribozyme-ncS3 complex, shows a slower migrating band with a slightly decreased FRET ratio (green-shifted). This decrease in FRET efficiency upon formation of the ribozyme D1-AP76-ncS3 complex is consistent with our solution assays (Figure 5a). These findings further support the idea that the precursor conformation is structurally distinct from the postcleavage ribozyme complex. Time-ResolVed FRET ReVeals the Global Architectures of the Precursor and Product Complexes. To quantify the global

structural changes in the trans-acting G76AP mutant ribozyme upon substrate cleavage, we used time-resolved FRET (trFRET) to measure the donor-acceptor distances in the FRET-labeled ribozyme and its substrate and 3′-product complexes as previously described (18). To this end, we recorded the time-resolved donor (fluorescein) decay curves of the ribozyme (D1-AP76), ribozyme-noncleavable substrate complex (D1-AP76-ncS3), and ribozyme-3′-product (D1AP76-3′P) complex, each of them singly labeled with donor as well as doubly labeled with donor and acceptor. The fluorescence decay curves of the donor-only-labeled complexes are nearly identical, and all show mean lifetimes of ∼4 ns (data not shown). We have shown previously that the anisotropies of the terminally attached donor and acceptor fluorophores in all HDV ribozyme complexes are low and similar in magnitude (18). This suggests that a decrease in donor lifetime in the doubly labeled complex is due to FRET and can be used to calculate donor-acceptor distances. Our tr-FRET data analysis (see Materials and Methods) confirms our results from steady-state FRET in solution and in gels, indicating that there are significant differences in the global architectures, particularly between the ribozyme-substrate and ribozyme-3′-product complexes. In particular, in the free D1-AP76 mutant ribozyme, the donor-acceptor distance distribution is centered around a mean distance of 48 Å (full width at half-maximum, fwhm, of 25 Å), in the D1-AP76ncS3 complex around 52 Å (fwhm of 21 Å), and in the D1AP76-3′P complex around 65 Å (fwhm of 33 Å) (Table 3). The full width at half-maximum, which is a measure of structural flexibility, suggests that a relatively rigid conformation is acquired prior to cleavage, while the ribozyme3′-product complex adopts a more flexible conformation upon product formation. Similar results were obtained when the measurements were carried out at 25 °C rather than under standard conditions at 37 °C (data not shown). These results confirm that our preparations are indeed structurally homogeneous, since in all three cases a single distance distribution fits the decay data well, as judged by the residuals and the reduced χ2 values (